Study design. This retrospective study analyzes the influence of lumber rotation on fusion extent in King type III AIS treatment by CD instrumentation. Objectives. To establish a guideline for fusion in King type III AIS Summary Study of background data. Rotational characteristics of the lumber curve may significantly affect the postoperativ e behavior of uninstrumented lumbar curve thus calling for a different lumbar rotation. Methods. Sixteen King type III AIS treated with CD were divided into two groups by the direction of lumbar curve rotation. It was opposite direction (OD) to the rotation of the major curve in 9 and same direction (SD) in 7. In OD, 6 were treated by selective thoracic fusion (TF) and 3 were fused to the stable vertebra (SV). In SD, 6 were treated by TF and 1 was fused to the SV. They were evaluated for balance, major and fractional curve correction after a minimum follow up of 2 years. Result. In OD, all curves were balanced regardless of the fusion extent with satisfactory curve correction. In SD-TF, all were clinically balanced with major curve correction of 75%, but all the fractional curve were overcorrected, adding on the major curve. In SD-SV, the curve was balanced with stable lumbar curve. Conclusion. In King type III curve with lumbar curve rotated to the opposite direction, selective thoracic fusion is sufficient. However, when the fractional curve is rotated in the same direction, fusion to the stable vertebra may be a safer choice.